A-Mem: Agentic Memory for LLM Agents
Open accessThe DOI matches its Crossref record; 33 of the 33 references it cites were found in a registry or an index; 11 of the 17 sources its claims cite were checked for support.
Abstract
This paper proposes a novel agentic memory system for LLM agents that can dynamically organize memories in an agentic way following Zettelkasten note-taking principles with dynamic indexing and linking.
Overview
CoChat Summary
TL;DR
A-MEM is an agentic memory system for LLM agents that, inspired by the Zettelkasten method, builds structured notes with LLM-generated keywords, tags, and contextual descriptions, then autonomously links and evolves them as new memories arrive. Across six foundation models on LoCoMo and DialSim, it outperforms LoCoMo, ReadAgent, MemoryBank, and MemGPT while using far fewer tokens.
Key findings
- On LoCoMo with GPT-4o-mini, A-MEM reaches 27.02 F1 / 20.09 BLEU-1 on Multi Hop and 45.85 / 36.67 on Temporal, versus MemGPT's 26.65 / 17.72 and 25.52 / 19.44; A-MEM's average ranking is 1.2 versus 2.4 for both LoCoMo and MemGPT.
- On DialSim, A-MEM scores F1 3.45, BLEU-1 3.37, ROUGE-L 3.54, ROUGE-2 3.60, METEOR 2.05, SBERT 19.51, versus LoCoMo (2.55, 3.13, 2.75, 0.90, 1.64, 15.76) and MemGPT (1.18, 1.07, 0.96, 0.42, 0.95, 8.54) — a 35% F1 improvement over LoCoMo and 192% over MemGPT.
- Ablation with GPT-4o-mini: removing both link generation and memory evolution drops Multi Hop F1 to 9.65; removing only memory evolution gives 21.35; full A-MEM gives 27.02.
- Cost/scaling: ~1,200 tokens per memory operation (85–93% reduction vs. ~16,900 for LoCoMo/MemGPT), under $0.0003 per operation, 5.4 s average with GPT-4o-mini and 1.1 s with local Llama 3.2 1B; retrieval time grows only from 0.31 µs to 3.70 µs from 1,000 to 1,000,000 memories.
Methods
- Design: each memory note is `{content, timestamp, keywords, tags, context, embedding, links}`; an LLM generates keywords/tags/context, a text encoder (all-minilm-l6-v2) produces embeddings, top-k nearest neighbors are retrieved by cosine similarity, and an LLM decides link generation and memory evolution (updating neighbors' context, keywords, tags).
- Data: LoCoMo (7,512 QA pairs, dialogues averaging 9K tokens over up to 35 sessions; single-hop, multi-hop, temporal, open-domain, adversarial) and DialSim (multi-party dialogues from Friends, The Big Bang Theory, The Office; 1,300 sessions, ~350,000 tokens, >1,000 questions per session).
- Evaluation: F1 and BLEU-1 as primary metrics, plus ROUGE-L, ROUGE-2, METEOR, and SBERT similarity; baselines LoCoMo, ReadAgent, MemoryBank, MemGPT; six foundation models (GPT-4o-mini, GPT-4o, Qwen2.5-1.5B/3B, Llama 3.2 1B/3B) plus DeepSeek-R1-32B, Claude 3.0 Haiku, and Claude 3.5 Haiku in the appendix; top-k retrieval with k=10 default and category-specific tuning.
Limitations
- Stated by the paper: memory organization quality may still depend on the underlying LLM's capabilities, and different LLMs may generate different contextual descriptions or connections; the implementation is text-only, with multimodal (image/audio) extension left to future work.
- Evident from the text: no error bars or significance tests are reported (the checklist answers "No" to statistical significance, citing API cost); the main-text results are reported for a single dataset family of long-term conversation QA, and the paper's own checklist notes broader societal impacts are not discussed.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
5 Valid · 0 Not valid · 3 Not checked · 55 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“These systems require agent developers to predefine memory storage structures, specify storage points within the workflow, and establish retrieval timing.”
What the sources say
- Partly supportedcites [25] [25] Charles Packer, Sarah Wooders, Kevin Lin, Vivian Fang, Shishir G Patil, Ion Stoica, and Joseph E Gonzalez. Memgpt: Towards llms as operating systems. arXiv preprint arXiv:2310.08560, 2023. sourceExperimental — accuracy not yet measured
Section 3.2.1 (p. 6) says the fixed-context baselines fetch documents independently of LLM inference, supporting the idea that retrieval is set by the workflow rather than chosen by the agent. Section 2.3 (p. 3) contrasts this with MemGPT’s self-directed memory edits and retrieval. The paper does not establish that existing memory systems generally require developers to predefine storage structures or specify storage points. (LLM, conf=0.78)
- Could not verifycites [39] [39] Wanjun Zhong, Lianghong Guo, Qiqi Gao, He Ye, and Yanlin Wang. Memorybank: Enhancing large language models with long-term memory. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 19724–19731, 2024. 12 Contents sourceExperimental — accuracy not yet measured
Sections 2.1 and 2.2 describe MemoryBank’s storage structure and retrieval mechanism, but do not say that existing systems require agent developers to predefine structures, choose workflow storage points, or set retrieval timing. Searches for those developer requirements found no passage addressing them. (LLM, conf=0.73)
- Not matchedcites [28] [28] Aymeric Roucher, Albert Villanova del Moral, Thomas Wolf, Leandro von Werra, and Erik Kaunismäki. 'smolagents': a smol library to build great agentic systems. https://github. com/huggingface/smolagents, 2025.
Its entry has no title and year that we can search by.
- Could not verifycites [21] [21] Zhiwei Liu, Weiran Yao, Jianguo Zhang, Liangwei Yang, Zuxin Liu, Juntao Tan, Prafulla K Choubey, Tian Lan, Jason Wu, Huan Wang, et al. Agentlite: A lightweight library for building and advancing task-oriented llm agent system. arXiv preprint arXiv:2402.15538, 2024. sourceExperimental — accuracy not yet measured
Section 2.1 describes when AgentLite retrieves an action-observation history and stores an observation; the Figure 1 description in Section 6 also describes its memory workflow. Neither passage says that existing memory systems require developers to predefine storage structures, storage points, or retrieval timing. Searches for those requirements found no passage addressing them. (LLM, conf=0.72)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Existing memory systems [25, 39, 28, 21] for LLM agents provide basic memory storage functionality.”
What the sources say
- Supportedcites [25] [25] Charles Packer, Sarah Wooders, Kevin Lin, Vivian Fang, Shishir G Patil, Ion Stoica, and Joseph E Gonzalez. Memgpt: Towards llms as operating systems. arXiv preprint arXiv:2310.08560, 2023. source
MemGPT is an LLM-agent memory system that stores information outside the context window. Section 2 (p. 2) describes its external context and memory-management functions, and the Figure 3 description (p. 3) identifies archival and recall storage databases. The claim does not say these systems provide only basic storage. (LLM, conf=0.88)
- Supportedcites [39] [39] Wanjun Zhong, Lianghong Guo, Qiqi Gao, He Ye, and Yanlin Wang. Memorybank: Enhancing large language models with long-term memory. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 19724–19731, 2024. 12 Contents source
MemoryBank is an LLM memory system with a memory-storage component. Section 2 describes that component as its primary data repository, and §2.1 says it stores conversation records, event summaries, and user information. This supports citing MemoryBank as an example of a system that provides memory storage functionality. (LLM, conf=0.91)
- Not matchedcites [28] [28] Aymeric Roucher, Albert Villanova del Moral, Thomas Wolf, Leandro von Werra, and Erik Kaunismäki. 'smolagents': a smol library to build great agentic systems. https://github. com/huggingface/smolagents, 2025.
Its entry has no title and year that we can search by.
- Supportedcites [21] [21] Zhiwei Liu, Weiran Yao, Jianguo Zhang, Liangwei Yang, Zuxin Liu, Juntao Tan, Prafulla K Choubey, Tian Lan, Jason Wu, Huan Wang, et al. Agentlite: A lightweight library for building and advancing task-oriented llm agent system. arXiv preprint arXiv:2402.15538, 2024. source
Section 2.1, “Individual Agent” (page not available), describes AgentLite’s Memory component as storing an agent’s action-observation chains and saving new actions and observations during execution. This supports citing AgentLite as an example of an LLM-agent system with basic memory storage functionality. (LLM, conf=0.90)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“While previous datasets contain dialogues with around 1K tokens over 4-5 sessions, LoCoMo features much longer conversations averaging 9K tokens spanning up to 35 sessions”
§ 4 Experiment > 4.1 Dataset and Evaluation
What the sources say
- Supportedcites [22] [22] Adyasha Maharana, Dong-Ho Lee, Sergey Tulyakov, Mohit Bansal, Francesco Barbieri, and Yuwei Fang. Evaluating very long-term conversational memory of llm agents. arXiv preprint arXiv:2402.17753, 2024. source
The Introduction (p. 2) reports that LoCoMo conversations average 9K tokens and span up to 35 sessions. Its comparison table (p. 2) lists earlier multi-session datasets at about 1,055–1,226 tokens across four or five sessions, supporting the stated contrast. (LLM, conf=0.94)
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“T-SNE Visualization of Memory Embeddings Showing More Organized Distribution with A-MEM (blue) Compared to Base Memory (red) Across Different Dialogues.”
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“with each memory operation costing less than $0.0003 when using commercial API services”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
25 more claims
Create a free account to see all 63 claims and every source verdict.
- Probably not a substantial claim. 16
- It may state two separate claims as one. 20
- The quote only partly states this claim. 2
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source doesn’t support the claim as written.
In the citing paper
“Relative to the O (1) dictionary lookup used by all prior systems [4], [5], [6], [7], [8], [9], lineage-gated retrieval trades at most 13 . 82 µ s in absolute terms for the safety guarantee of Theorem 1-a tradeoff Experiments 1-2 show buys elimination of an 18 . 8 -25 . 5% structural leak rate at a 13 -14 percentage-point reuse cost.”
From Lineage-Aware Memory Governance: A Derivation-Gated Framework for Privacy-Preserving Column-Level Access Control in Enterprise AI Agents (2026) · cites this paper as [6]
Checked by a reasoning model — may be wrong.
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“This emerging class of systems extends the capabilities of traditional AI Agents by enabling multiple intelligent entities to collaboratively pursue goals through structured communication [144-146], shared memory [147,148], and dynamic role assignment [22].”
From AI Agents vs. Agentic AI: A Conceptual taxonomy, applications and challenges (2026) · cites this paper as [147]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“from Wang et al. [55], Xu et al. [45] build memory to improve long context performance for specific workflows which suggest that diverse evaluation workflows are needed for better testing of generalizable memory components.”
From MEMTRACK: Evaluating Long-Term Memory and State Tracking in Multi-Platform Dynamic Agent Environments (2025) · cites this paper as [45]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Some other approaches introduce external memory modules ( e.g., summarizers or retrievers) [63, 29, 13, 57], but these are typically trained separately and cannot be optimized end-to-end with the agent's policy. This also introduces additional engineering overhead, as engineers must manage and integrate two separate models.”
From MEM1: Learning to Synergize Memory and Reasoning for Efficient Long-Horizon Agents (2025) · cites this paper as [57]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Some others, such as HiAgent[5], A-Mem[23], MemAgent [28] use hierarchical designs to separate working buffers from episodic and long-term stores to relieve context pressure and improve persistence.”
From Learning Hierarchical Procedural Memory for LLM Agents through Bayesian Selection and Contrastive Refinement (2025) · cites this paper as [23]
Experimental — accuracy not yet measured
9 more claims about this paper
Create a free account to see every claim other papers make about this one.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
§ APPENDIX
“Using GPT-4o-mini, A-MEM achieves a ROUGE-L score of 44.27 in Multi-Hop tasks, more than doubling LoComo's 18.09.”
§ APPENDIX
“LOCOMO 9.64 23.92 2.01 18.09 3.40 11.58 26.48 40.20 60.46 69.59 READAGENT 2.47 9.45 0.95 13.12 0.55 5.76 2.99 9.92 6.66 9.79 MEMORYBANK 1.18 5.43 0.52 9.64 0.97 5.77 1.64 6.63 4.55 7.35 MEMGPT 10.58 25.60 4.76 25.22 0.76 9.14 28.44 42.24 36.62 43.75 A-MEM 10.61 25.86 21.39 44.27 3.42 12.09 29.50 45.18 42.62 50.04”
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
§ APPENDIX
“The Multi-Hop category showcases particularly striking results, where Qwen2.5-15b with A-MEM achieves a ROUGE-L score of 27.23, dramatically surpassing LoComo's 4.68 and ReadAgent's 2.81 - representing a nearly six-fold improvement.”
§ APPENDIX
“Multi Hop Temporal Open Domain Single Hop Adversial RGE-2 RGE-L RGE-2 RGE-L RGE-2 RGE-L RGE-2 RGE-L RGE-2 RGE-L”
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
§ APPENDIX
“This significant advantage maintains consistency across other metrics, with METEOR scores of 23.43 versus 7.61 and SBERT scores of 70.49 versus 52.30.”
§ APPENDIX
“Multi Hop Temporal Open Domain Single Hop Adversial ME SBERT ME SBERT ME SBERT ME SBERT ME SBERT”
Rated major: it would touch a result — a value a result rests on, reported and recomputed p-values on opposite sides of the significance level, two passages that conflict with no sign which is right, or a cited work retracted before this paper was published.
In the paper
§ 4 Experiment > 4.3 Empricial Results
“our A-MEM demonstrates superior performance in Multi-Hop tasks achieves at least two times better performance”
§ 4 Experiment > 4.1 Dataset and Evaluation
“MEMGPT 26.65 17.72 25.52 19.44 9.15 7.44 41.04 34.34 43.29 42.73 2.4 2.4 16,977 A-MEM 27.02 20.09 45.85 36.67 12.14 12.00 44.65 37.06 50.03 49.47 1.2 1.2 2,520”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 10 20 30 40 50
“44.55 25.43 28.31 32.12 34.32 37.02”
§ 4 Experiment > 4.1 Dataset and Evaluation
“A-MEM 27.02 20.09 45.85 36.67 12.14 12.00 44.65 37.06 50.03 49.47 1.2 1.2 2,520”
Citations
Cites
The publisher hasn’t deposited this paper’s reference list.
Research works printed but not linked: 4
- [1][1] Sönke Ahrens. How to Take Smart Notes: One Simple Technique to Boost Writing, Learning and Thinking. Amazon, 2017. Second Edition.
- [5][5] Satanjeev Banerjee and Alon Lavie. Meteor: An automatic metric for mt evaluation with improved correlation with human judgments. In Proceedings of the acl workshop on intrinsic and extrinsic evaluation measures for machine translation and/or summarization, pages 65–72, 2005.
- [18][18] Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, et al. Retrieval-augmented generation for knowledge-intensive nlp tasks. Advances in Neural Information Processing Systems, 33:9459–9474, 2020.
- [19][19] Chin-Yew Lin. Rouge: A package for automatic evaluation of summaries. In Text summarization branches out, pages 74–81, 2004.
Cited by
Items 701–800 of 1,030
Liu, Q., Wang, M., & Xie, J. (2026) · 2026 6th International Conference on Machine Learning and Intelligent Systems Engineering (MLISE) · doi:10.1109/mlise70044.2026.11607496
“A-Mem [12] and related work likewise focus more on organization and retrieval than content-level write control.”
From Failed Trajectories to Reliable LLM Agents: Diagnosing and Repairing Harness Flaws
Chen, M., Wang, J., Liu, Z., Wang, Y., Zheng, H., & Wang, Q. (2026) · arXiv · doi:10.48550/arxiv.2606.06324
“…skills from bad cases [52], ReCreate creates domain agent scaffolds from experience [7], Agent Workflow Memory induces reusable workflows [53], and Agent KB, A-MEM, MemInsight, ACE, and GEPA study strategy stores, dynamic memory, evolving context, and prompt evolution [54], [55], [56], [57], [38].”
MemSkill: Learning and Evolving Memory Skills for Self-Evolving Agents
Zhang, H., Long, Q., Bao, J., Feng, T., Zhang, W., Yue, H., & Wang, W. (2026) · arXiv · doi:10.48550/arxiv.2602.02474
“This need has driven rapid progress in agent memory, including approaches that summarize and retrieve past interactions or manage external memory stores (Kang et al., 2025; Chhikara et al., 2025; Packer et al., 2023; Xu et al., 2025).”
Agent KB: Leveraging Cross-Domain Experience for Agentic Problem Solving
Tang, X., Qin, T., Peng, T., Zhou, Z., Shao, D., Du, T., Wei, X., Xia, P., Wu, F., Zhu, H., et al. (2025) · arXiv · doi:10.48550/arxiv.2507.06229
Mem-α: Learning Memory Construction via Reinforcement Learning
Wang, Y., Takanobu, R., Liang, Z., Mao, Y., Hu, Y., McAuley, J., & Wu, X. (2025) · arXiv · doi:10.48550/arxiv.2509.25911
“Second, they require direct access to model internals, making them incompatible with proprietary systems (e.g., GPT-4 LLM Agents with External Memory An alternative approach equips language models with ex-ternal memory systems built on databases or vector stores (Zhang et al., 2025a), as demonstrated by MemGAS (Xu et…”
Self-citationJiang, B., Yuan, Y., Shen, M., Hao, Z., Xu, Z., Chen, Z., Liu, Z., Vijjini, A. R., He, J., Yu, H., et al. (2025) · arXiv · doi:10.48550/arxiv.2512.06688
“Agentic memory (Yu et al., 2025; Xu et al., 2025; Zhong et al., 2024; Zhang et al., 2025d; Li et al., 2025b) offers a scalable mechanism for language models to distill long-term user information into a compact user persona.”
General Agentic Memory Via Deep Research
Yan, B. Y., Li, C., Qian, H., Lu, S., & Liu, Z. (2025) · arXiv · doi:10.48550/arxiv.2511.18423
“2) Memory-based methods , including A-Mem [8], Mem0 [9], MemoryOS [10] and LightMem [17].”
MemoryBench: A Benchmark for Memory and Continual Learning in LLM Systems
Ai, Q., Tang, Y., Wang, C., Long, J., Su, W., & Liu, Y. (2025) · arXiv · doi:10.48550/arxiv.2510.17281
“A typical LLM memory system support multiple operations such as writing(Zhong et al., 2024), summarizing (Liu et al., 2023), updating (Xu et al., 2025a), deleting (Park et al., 2023), and retrieval (Packer et al., 2023).”
Lin, Z., Hao, X., Fu, R., Cui, S., Chen, K., Li, C., Li, Z., & Xiong, F. (2026) · arXiv · doi:10.48550/arxiv.2604.16548
Two Heads are Better Than One: Test-time Scaling of Multi-agent Collaborative Reasoning
Jin, C., Peng, H., Zhang, Q., Zhou, Y., Tang, Y., Che, T., & Metaxas, D. N. (2025) · arXiv · doi:10.48550/arxiv.2504.09772
DeepWideSearch: Benchmarking Depth and Width in Agentic Information Seeking
Lan, T., Zhu, B., Jia, Q., Ren, J., Li, H., Wang, L., Xu, Z., Luo, W., & Zhang, K. (2025) · arXiv · doi:10.48550/arxiv.2510.20168
Westhäußer, R., Minker, W., & Zepf, S. (2025) · arXiv · doi:10.48550/arxiv.2510.07925
M2A: Multimodal Memory Agent with Dual-Layer Hybrid Memory for Long-Term Personalized Interactions
Feng, J., Xu, B., Chen, J., Dai, M., Wu, C., Li, H., Zeng, B., Xie, Y., Liang, H., Lu, M., et al. (2026) · arXiv · doi:10.48550/arxiv.2602.07624
“We use the A-MEM baseline.”
InfluentialFrom Experience to Strategy: Empowering LLM Agents with Trainable Graph Memory
Xia, S., Xu, Z., Chai, J., Fan, W., Song, Y., Wang, X., Yin, G., Lin, W., Zhang, H., & Wang, J. (2025) · arXiv · doi:10.48550/arxiv.2511.07800
Liu, S., Zhu, J., Shu, L., Lin, J., Chen, Y., Zhang, H., Zhang, C., Xu, D., Li, J., Tang, B., et al. (2026) · arXiv · doi:10.48550/arxiv.2603.23231
“Despite these advancements, it remains designed for “Needle-in-a-Haystack” tasks, in which isolated preferences are retrieved from vast volumes of unrelated context.”
Intrinsic Memory Agents: Heterogeneous Multi-Agent LLM Systems through Structured Contextual Memory
Yuen, S., Medina, F. G., Su, T., Du, Y., & Sobey, A. J. (2025) · arXiv · doi:10.48550/arxiv.2508.08997
“…solutions such as Retrieval-Augmented Generation (RAG) (Lewis et al. 2020; Gao et al. 2024) and agentic memory approaches (Packer et al. 2024; Xu et al. 2025; Chhikara et al. 2025) are designed for single-agent and user interaction scenarios, which do not account for the volume of…”
Mem-T: Densifying Rewards for Long-Horizon Memory Agents
Yue, Y., Peng, B., Fan, X., Guo, J., Li, Q., & Zhang, Y. (2026) · arXiv · doi:10.48550/arxiv.2601.23014
Self-citationMishra, S., Niroula, S., Yadav, U., Thakur, D., Gyawali, S., & Gaire, S. (2026) · arXiv · doi:10.48550/arxiv.2603.07379
AMemGym: Interactive Memory Benchmarking for Assistants in Long-Horizon Conversations
Jiayang, C., Ru, D., Qiu, L., Li, Y., Cao, X., Song, Y., & Cai, X. (2026) · arXiv · doi:10.48550/arxiv.2603.01966
“Beyond these foundational implementations, we extend our evaluation to include established memory agent frameworks, such as Mem0-G (Chhikara et al., 2025), Nemori (Nan et al., 2025), and A-Mem (Xu et al., 2025).”
AgentArch: A Comprehensive Benchmark to Evaluate Agent Architectures in Enterprise
Bogavelli, T., Sharma, R., & Subramani, H. (2025) · arXiv · doi:10.48550/arxiv.2509.10769
“…a self directed operating system with different storage tiers (Packer et al. 2024) , creating dynamic knowledge networks for connecting information (Xu et al. 2025), and implementing a dual-track memory system meant to resemble long-term and short-term memory in the human brain (Liu et al. 2024a).”
Recursive Experiential-Working Memory Evolution for Long-Horizon Agent Harnesses
Yu, Z., Wu, Y., Yin, Z., Chen, K., Zhao, Z., Wang, M., Yan, S., & Yang, L. (2026) · arXiv · doi:10.48550/arxiv.2608.24876
“…is the memory itself: experience is written back after each task and maintained through add, update and delete operations (Chhikara et al., 2025; Xu et al., 2025), distilled into reusable strategies (Ouyang et al., 2026), or repaired from failed runs (Zhang et al., 2026a; Zhu et al., 2025).”
Lu, Y., Cheng, W., Zhang, Z., & Tang, H. (2026) · arXiv · doi:10.48550/arxiv.2602.16493
“Other approaches compress or synthesize memory representations to reduce long-horizon overhead (Zhou et al., 2025; Zhang et al., 2025a) or organize memories into evolving networks for indexing and updates (Xu et al., 2025).”
Coding Agents are Effective Long-Context Processors
Cao, W., Yin, X., Dhingra, B., & Zhou, S. (2026) · arXiv · doi:10.48550/arxiv.2603.20432
“Agent with Long-Term Memory A growing body of work has focused on building and optimizing memory-centric agentic architectures through various memory manipulation strategies (Chhikara et al., 2025; Hu et al., 2026; Huo et al., 2026; Xu et al., 2025).”
SYNAPSE: Empowering LLM Agents with Episodic-Semantic Memory via Spreading Activation
Jiang, H., Chen, J., Pan, Y., Chen, L., You, W., Zhou, Y., Zhang, R., Sikora, A., Zhao, L., Abate, Y., et al. (2026) · arXiv
“A-Mem (Xu et al., 2025) and AriGraph (Anokhin et al., 2025) utilize LLMs to maintain dynamic knowledge graphs, while Hip-poRAG (Gutiérrez et al., 2024) adapts Personal PageRank for retrieval.”
AgentLongBench: A Controllable Long Benchmark For Long-Contexts Agents via Environment Rollouts
Fang, S., Wang, Y., Liu, X., Lu, J., Tan, C., Chen, X., Zheng, Y., Huang, X., & Qiu, X. (2026) · arXiv · doi:10.48550/arxiv.2601.20730
“We implement standard RAG [35] alongside specialized agentic memory systems such as A-Mem [68], Mem0 [11], and Mem-oryOS [30].”
MemoBrain: Executive Memory as an Agentic Brain for Reasoning
Qian, H., Cao, Z., & Liu, Z. (2026) · arXiv · doi:10.48550/arxiv.2601.08079
“…to retain and organize information over long interactions, such as long-term dialogue or large collections of documents, and is commonly instantiated through summarization, parametric storage, or graph-based representations (Xu et al., 2025; Yan et al., 2025a; Yu et al., 2025; Kang et al., 2025).”
When Agents Remember Too Much: Memory Poisoning Attacks on Large Language Model Agents
Torres, G., Shrestha, S., & Misra, S. (2026) · arXiv · doi:10.48550/arxiv.2607.06595
“Given the absence of long-term memory agents that are purpose-built for a personal assistant use case, we selected five state-of-the-art agent architectures and adapted them to a common evaluation setting: A-Mem [22], Mem0 [32], ExpeL [4], Letta (formerly MemGPT) [3], [33], and Memo-ryOS [5].”
SKILL.nb: Selective Formalization and Gated Execution for Durable Agent Workflows
Hattami, A. E., Chapados, N., & Pal, C. (2026) · arXiv · doi:10.48550/arxiv.2606.08049
PreAct: Computer-Using Agents that Get Faster on Repeated Tasks
Li, B. (2026) · arXiv · doi:10.48550/arxiv.2606.17929
“More broadly, agent-memory systems store and retrieve past interactions: generative-agent memory streams [33], operating-system-style memory [32], and production memory stacks [9, 55] (surveyed in [62]), typically consulted by retrieval-augmented generation [21].”
Mesh Memory Protocol: Semantic Infrastructure for Multi-Agent LLM Systems
Xu, H. (2026) · arXiv · doi:10.48550/arxiv.2604.19540
“A-MEM (Xu, W. et al. 2025) performs write-time memory organisation — generating structured notes and linking to prior memories as each entry is added.”
AgentKVShift: Efficient KV Cache Reuse for Agentic Memory Systems
Pandey, N. P., Kong, J., Hu, L., Zhao, Q., Zhao, Y., Gungor, O., Zhang, H., & Rosing, T. (2026) · arXiv · doi:10.48550/arxiv.2607.21604
“Note-based systems [24] generate Zettelkasten-style notes enriched with summaries, keywords, and tags, building on a longer line of work in dialogue compression and fact-extraction memory that distill interactions into compact, self-contained units [9, 30, 16].”
TraceMem: Weaving Narrative Memory Schemata from User Conversational Traces
Shu, Y., Liu, P., Zhang, T., Gao, R., Ma, J., & Sun, C. (2026) · arXiv · doi:10.48550/arxiv.2602.09712
“Moving forward, cognitively-inspired approaches emerge as a promising direction, exemplified by recent works like A-Mem [34] and Nemori [23].”
Potapov, A. S., & Potapova, V. (2025) · Lecture notes in computer science · doi:10.1007/978-3-032-00800-8_11 · Cited by 2
Liu, J., Li, Y., Yu, Z., Su, S.-C., Gao, S., & Huang, Y. (2025) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2025.findings-emnlp.101
“This capability stems from their extensive parameterization and sophisticated deep neural architectures, which enable effective extraction and modeling of high-frequency data patterns during the pre-training phase (Xu et al., 2025; Zhao et al., 2024; Kim et al., 2024).”
A Novel Self-Evolution Framework for Large Language ModelsPreprint
Sun, H., Zhang, Z., & Zeng, S. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2507.15281
The Agentic AI Security Adoption Matrix: Understanding Readiness and Resistance Across DomainsPreprint
WAIZEL, G. (2026) · Research Square · doi:10.21203/rs.3.rs-8480906/v1 · Cited by 1
A Survey on Generative Knowledge Graph Construction
洪, 钊. (2026) · Artificial Intelligence and Robotics Research · doi:10.12677/airr.2026.151017
Towards Reliable ML Feature Engineering via Planning in Constrained-Topology of LLM AgentsPreprint
Thakur, H., Kamath, A., Muthyala, A., Sanmukhani, D., Mukund, S., & Katukuri, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2601.10820 · Cited by 1
“Real-world environments impose additional constraints: many agents are fixed and can only be modified via prompting, necessitating frequent manual tuning, while team workflows evolve rapidly, making updates harder [24, 32, 13, 23, 26, 11].”
RAG without Forgetting: Continual Query-Infused Key Memory
Hu, Y., Li, S., Ramakrishnan, N., & Zhao, L. (2026) · PubMed · doi:10.48550/arxiv.2602.05152 · Cited by 1
THE ERA OF SKILL GROWING AGENTS:SHIFT FROM TASK EXECUTION TO SKILL GROWTHPreprint
Li, Z., Chen, J., Jiang, Y., Han, Z., Huang, W., Shi, J., Zhang, J., He, J., Zhong, W., Zhang, D., et al. (2026) · doi:10.36227/techrxiv.177041975.54183859/v1
Not All Gap Correction Helps: A Geometric View of External Information in LLM InferencePreprint
Wang, D., Zhang, X., Xu, K., Zhu, Q., Che, W., & Deng, Y. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2602.08294 · Cited by 1
Wang, L., Liu, Y., Lu, J., Guo, X., Li, Z., & Huang, N. (2026) · ACM International Conference on Web Search and Data Mining (WSDM) · doi:10.1145/3773966.3777936 · Cited by 1
Sun, Y., Yu, H., Jiang, W., Yu, X., Zhan, D., Wang, L., Ren, S., Sun, Y., & Zhu, T. (2026) · Preprints.org · doi:10.20944/preprints202602.1655.v1 · Cited by 2
Towards Autonomous Memory AgentsPreprint
Wu, X., Zhang, R., Hussain, M. A., & Lu, Y. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2602.22406 · Cited by 1
“This has driven the recent interest in memory agents [3, 4, 24, 40, 45]: instead of updating model weights, the agent extracts critical experiences from the conversation history, user preferences, task outcomes, domain knowledge, and reasoning patterns, and puts them into an external memory store…”
LLM Agent Memory: A Survey from a Unified Representation–Management PerspectivePreprint
Tang, Z., He, X., Zhao, T., Wei, F., Liu, X., Dong, P., (32718), Q. W., Li, Q., Wang, H., Chen, R., et al. (2026) · Preprints.org · doi:10.20944/preprints202603.0359.v1 · Cited by 2
Chen, B., Gao, Q., Wan, S., Yan, R., Sun, S., li, X., Liu, T., Liu, T., & Zhao, L. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2603.05873
“Recent progress on AI agents and agentic memory offers a new promising paradigm [30,3,27].”
Evolving Generalist Virtual Agents with Generative and Associative Memory
Zhang, Z., Bu, W., Pan, K., Miao, B., Zhang, W., Wang, G., Ji, W., Tang, R., Li, J., & Tang, S. (2026) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v40i15.38300 · Cited by 6
“This fragmented storage and brittle retrieval process hinders agents from reusing and composing knowledge from experiences that are contextually relevant but semantically distant, thereby limiting their effectiveness in open-world environments (Xu et al. 2025a).”
Personal Intelligence: Toward a User-Governed Preference Substrate for the Age of Agentic AIPreprint
Montes, G. A. (2026) · Preprints.org · doi:10.20944/preprints202603.1627.v1
EACC: Efficient Agent Context Cache Sharing for Multi-Agent Systems
Cheng, S., Gu, Y., Wu, C., & Meng, J. (2026) · Lecture notes in computer science · doi:10.1007/978-981-95-8405-5_9
Aditto, A. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.00009
“A-MEM [Xu et al., 2025] introduces agentic memory with self-organizing Zettelkasten structure.”
ACF: A Collaborative Framework for Agent Covert Communication under Cognitive AsymmetryPreprint
Wu, W., Huang, K., Wei, Y. K., Yang, Z., & Zhou, L. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.08276
“…matching, causing channel failure. b) Cognitive Asymmetry in Agents: Autonomous agents maintain a dynamic cognitive state K t = K pub t ∥ K priv t (where ∥ denotes context concatenation), comprising observable public history ( K ) and localized private memory ( K ) [20], [21], [29], [30].”
From Speech to Profile: A Protocol-Driven LLM Agent for Psychological Profile GenerationPreprint
Yang, X., Yang, Y., Guo, Z., Zhou, Y., Yan, N., & Wang, L. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.10161
“Retrieval-augmented generation (RAG) and agent memory techniques offer potential remedies by grounding generation in retrieved context[14, 15, 16, 17].”
Wen, S., & Ku, B. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.11243
Contexty: Capturing and Organizing In-situ Thoughts for Context-Aware AI Support
Kim, Y., Park, C., Son, K., Yang, S., & Kim, J. (2026) · ACM Symposium on User Interface Software and Technology (UIST) · doi:10.1145/3830398.3830663
Gadzhiev, A., & Kislov, A. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.11563
“Every published LoCoMo baseline—TiMem, Mem0, MemOS, MemoryOS, A-MEM, LangMem—treats conversation as a retrieval problem over raw or lightly summarized dialogue segments.”
InfluentialCooperative Memory Paging with Keyword Bookmarks for Long-Horizon LLM ConversationsPreprint
Liu, Z. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.12376
“Mem0 (Chhikara et al., 2025) and A-MEM (Xu et al., 2025) build pro-Table et al., 2024) adds long-term memory with summarization and forgetting for multi-turn conversations, while SeCom (Pan et al., 2025) studies memory granularity and segmentation for personalized conversational agents.”
vstash: Local-First Hybrid Retrieval with Adaptive Fusion for LLM AgentsPreprint
Steffens, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.15484
“A-MEM (Xu et al., 2025) uses agentic self-organization.”
Retention Consequence in Lifecycle Memory ControlPreprint
Han, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.16774
“Mem-oryBank, Mem0, A-MEM, All-Mem, AtomMem, and reflective memory-management approaches study how memories are extracted, updated, consolidated, retrieved, and revised over time (Zhong et al., 2023; Chhikara et al., 2025; Xu et al., 2025; Lv et al., 2026; Huo et al., 2026; Tan et al., 2025b).”
Four-Axis Decision Alignment for Long-Horizon Enterprise AI AgentsPreprint
Srininvasan, V. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.19457
“HyMem [2], TiMem [3], H-MEM [4], GAM [5], MemGPT/Letta [6], MIRIX [7], MEM1 [8], and A-Mem [9] each propose a memory mechanism for long-horizon agents.”
Stateless Decision Memory for Enterprise AI AgentsPreprint
Srinivasan, V. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.20158
“A-Mem [9] uses agentic memory operators.”
Xu, N. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.20598
“A-MEM [8] draws on the Zettelkasten method to give LLM agents a self-organizing note system: when a new memory is added, the agent generates structured attributes (keywords, tags, context) and the system dynamically links it to existing notes, occasionally rewriting those notes’ representations.”
PragLocker: Protecting Agent Intellectual Property in Untrusted Deployments via Non-Portable PromptsPreprint
Li, Q., Bao, Y., Hu, J., Zhang, W., Chen, J., Zhu, H., Jin, Y., & Zhang, X. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.05974
“(Lee et al., 2024) (long-context reading with episodic memory and gist-level compression), and A-MEM (Xu et al., 2025) (long-term memory via structured notes and dynamic indexing).”
Improving Multi-turn Dialogue Consistency with Self-Recall Thinking
Pang, R., Lan, T., Liu, L., Huang, X., Tong, P., & Zhang, X. (2026) · Lecture notes in computer science · doi:10.1007/978-981-92-0372-7_5
“A-MEM [32] adopts agentic organization, dynamically indexing and linking memory units as new evidence arrives.”
Jiang, S., Wang, Y., & Yuan, X. (2026) · Expert Systems with Applications · doi:10.1016/j.eswa.2026.132623
MemReread: Enhancing Agentic Long-Context Reasoning via Memory-Guided RereadingPreprint
Ji, B., Weng, X., Li, J., Tang, Z., Lou, Y., & Zhang, M. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.10268 · Cited by 1
“As shown in Figure 7b, our method outperforms GRPO baselines across all benchmarks and adaptively modulates reading passes according to task complexity.”
CogniFold: Always-On Proactive Memory via Cognitive FoldingPreprint
Wang, S., Duan, Y., Deng, Y., Zhao, R., Shi, D., Deng, M., Chen, C., Wang, Y., & Zhou, X. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.13438
“…is a substrate for conceptual bootstrapping [5, 66]: recursively scaffolding higher-level cognition from accumulated structure—a transparent, auditable form of test-time learning, distinct from both surface-level text rewriting (e.g., A-Mem [63]) and opaque gradient updates (e.g., Titans [2]).”
Emotion-Attended Stateful Memory (EASM):The Architecture for Hyper-Personalization at ScalePreprint
Kotecha, V., & Gupta, V. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.14833
Improving Multi-turn Dialogue Consistency with Self-Recall ThinkingPreprint
Pang, R., Lan, T., Liu, L., Huang, X., Tong, P., & Zhang, X. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.15102
“A-MEM [32] adopts agentic organization, dynamically indexing and linking memory units as new evidence arrives.”
Is One Score Enough? Rethinking the Evaluation of Sequentially Evolving LLM MemoryPreprint
Dong, S., Chen, Z., Shi, C., Wang, P., Li, J., & Shen, C. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.15384
“Recent memory-augmented agents extract reusable information from prior trajectories, feedback, reflections, or workflows, and use such information to improve future reasoning and decision-making [29, 6, 9, 46, 41].”
Toward Human-Aligned, Modular, and Emotion-Aware Multi-Agent Systems: A Systematic Literature ReviewPreprint
Ghanem, H., Cruz, C., & Cherifi, H. (2026) · Research Square · doi:10.21203/rs.3.rs-9485421/v1
Generating Expressive and Customizable Evals for Timeseries Data Analysis Agents with AgentFuel
Maddi, A., Naval, P. P., Mande, D., Girish, M., Duan, S., & Sekar, V. (2026) · ACM Conference on AI and Agentic Systems (CAIS) · doi:10.1145/3786335.3813135
EvoIR-Agent: Self-Evolving Image Restoration Agentic System via Experience-Driven LearningPreprint
Zhuang, K., Wu, J., & Jin, Z. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.22208 · Cited by 1
“Recent advancements [21, 22, 23, 24] have demonstrated that explicit experience drives self-evolving skills in general-purpose agents.”
Wang, X., Feng, T., Liu, H., & Zhu, W. (2026) · Springer eBooks · doi:10.1007/978-981-95-7750-7_5 · Cited by 1
MedExpMem: Adapting Experience Memory for Differential Diagnosis
Feng, Q., Huang, Z., Zhu, Y., Gu, Y., Chu, C. W., Zhang, X., & Dou, Q. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-38098-2_41
Hazoom, M., Patel, G., Talmor, A., & Hope, T. (2026) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/2026.emnlp-industry.196
Memory Architectures for Multi-Turn Text-to-SQL: A Benchmark and Empirical StudyPreprint
Tummalapenta, R. K., & Addanki, S. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.26394
“A-MEM (Xu et al., 2025) introduces agentic memory with Zettelkasten-inspired note construction and dynamic link generation over an embedding store.”
MGRetrieval: Memory-Guided Reflective Retrieval for Long-Term Dialogue AgentsPreprint
Wang, T., & Dong, Y. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.27437 · Cited by 1
Omelchenko, I., & Strukov, V. (2025) · Bulletin of V N Karazin Kharkiv National University series «Mathematical modeling Information technology Automated control systems» · doi:10.26565/2304-6201-2025-68-06
Hierarchical Long-Term Semantic Memory for LinkedIn's Hiring Agent
Xu, Z., Zhang, S., Poyraz, E., Li, Y., Jin, Y., Lu, X., Gu, X., Ramgopal, K., Bodigutla, P. K., & Wang, X. (2026) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3770855.3818432 · Cited by 3
“Agentic memory systems [4, 10, 26, 30, 31, 33] support an agent’s long-term interaction with its environment by persisting user-specific signals from historical data and leveraging them to personalize future agent–user exchanges.”
InfluentialGraph-Structured Persistent Memory for Efficient LLM-Based Computer Use Agents
Vorvul, D., Musienko, A. P., Galchenko, I., Myroniuk, M., & Собчук, А. (2026) · Axioms · doi:10.3390/axioms15060415
KACE: Knowledge-Adaptive Context Engineering for Mathematical ReasoningPreprint
Parashar, J., & Bhandarkar, S. M. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.00532
“…in working context is itself a learnable decision: Memory-as-Action [Zhang et al., 2025c] treats context curation as an explicit policy, A-MEM [Xu et al., 2025] and Mem0 [Chhikara et al., 2025] build production-oriented agentic memory stacks, and MemGPT [Packer et al., 2023] formalizes…”
Beyond Storage: Toward Living Memory Systems for LLMs on the Edge
Foley, M., Radu, V., Pegoraro, M., Pour, M. V., Faricy, F., & Girolami, M. (2026) · International Conference on Mobile Systems, Applications, and Services · doi:10.1145/3812836.3814755
CrossBench: Reproducible LLM Memory Evaluation Across Conversational and Coding Workloads
Radu, V., Foley, M., Pour, M. V., Pegoraro, M., Faricy, F., & Girolami, M. (2026) · International Conference on Mobile Systems, Applications, and Services · doi:10.1145/3812836.3814760
AVTrack: Audio-Visual Tracking in Human-centric Complex ScenesPreprint
Wang, Y., Yun, Z. H. O. U., Zhang, Z., & Ding, H. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.02724 · Cited by 2
RAMPART: Registry-based Agentic Memory with Priority-Aware Runtime TransformationPreprint
Tomczak, N. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.04628
“Xu et al. approached the same challenge from a memory organisation standpoint, introducing A-MEM, which structures agent memory as a dynamic graph of Zettelkasten-inspired linked notes and demonstrates competitive performance on long-term conversational benchmarks [19].”
Chen, J., Lu, Y., Su, J., Chen, C., Xiao, S., Zhang, Z. G., (32846), Y. W., Li, Y., Zhao, J., Wu, T., et al. (2026) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/2026.findings-emnlp.2171
Personalization Meets Safety:Mechanisms,Risks,and Mitigations in Personalized LLMsPreprint
Luo, Y., Han, X., Bai, R., Huang, X., Wang, Y., Hu, Q., Wang, Q., Zhao, C., Liu, J., Geng, C., et al. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.09038
Web2BigTable: A Bi-Level Multi-Agent LLM System for Internet-Scale Information Search and ExtractionPreprint
Fang, M., Huang, Y. Q., Chen, Y., He, Z., Chen, Y., Lee, K. Y., Zhou, H., Luo, W., & Wang, Y.-X. (2026) · Research Square · doi:10.21203/rs.3.rs-9930430/v1 · Cited by 1
RunAgent SuperBrowser: A Theory of Autonomous Web Navigation Grounded in Human Browsing BehaviourPreprint
Mostafa, R., & Saha, S. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.09399
One Token per Multimodal Evidence: Latent Memory for Resource-Constrained QAPreprint
Zheng, Z., Meng, Z., Luan, H., Liu, W., & Lee, W. S. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.10572
What Spatial Memory Must Store: Occlusion as the Test for Language-Agent MemoryPreprint
권도언, & Bang, J. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.10299
“A recent wave structures memory as a graph: A-MEM [3] builds an evolving Zettelkasten of linked notes; Mem0 [4] extracts and consolidates facts with an optional graph variant; Zep/Graphiti [5] maintains a temporal knowledge graph.”
Knowledge Graph Enhanced Memory-Augmented Retrieval for Long Context ModelingPreprint
Alselwi, G., Suleiman, B., Xue, H., Jameel, S., Hacid, H., Salim, F. D., & Razzak, I. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.14047
“Agentic systems like Mem0 (Chhikara et al., 2025) and A-Mem (Xu et al., 2025) extend RAG for interactive agents but focus on cross-session persistence rather than within-document structural reasoning.”
Token-Operations-Oriented Inference Optimization Techniques for Large ModelsPreprint
Lian, S., Wang, K., Liu, Z., Liu, W., Hua, M., Liu, Y., Yan, J., Wang, X., Wang, C., Zhang, Y., et al. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.20295
Self-citationA Self-Evolving LLM Agent Framework for Role-Based Norm Compliance in Healthcare
Ruan, H., Jiang, X., Zhanpeng, L. I., Jia, W., Xu, X., Shan, X.-F., Chen, S., & Ye, X. (2026) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2026.findings-acl.1133
“Structured memory contains (i) an experience buffer and (ii) an identity & norm state (Zhang et al., 2025; Hatalis et al., 2023; Xu et al., 2025).”
KnowMe-Bench: Benchmarking Person Understanding for Lifelong Digital Companions
Wu, T., Chen, Z., Weng, Z., Wang, S., Zhang, S., Hu, S., Wu, S., Lan, Q., Wang, H., & Chen, R. (2026) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2026.acl-long.1394 · Cited by 4
Demystify the Role of Memory in Machine Learning Engineering Agents
Zhao, X., Wang, J., Chen, Y., Pan, M., Yeh, C. M., Sun, J., Zheng, Y., Das, M., & Chen, T. (2026) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2026.findings-acl.525 · Cited by 1
Metadata
- Authors
- Xu, W., Liang, Z., Mei, K., Gao, H., Tan, J., & Zhang, Y.
- Affiliations
- Rutgers, The State University of New Jersey (US)
- Ant Group (China) (CN)
- Salesforce (United States) (US)
- Venue
- Advances in Neural Information Processing Systems 38
- Published
- 2025
- Pages
- 20004-20031
- ISSN
- 1049-5258 (ISSN-L)
- OpenAlex
- W4407759090
- Cited by
- 51
- References
- 0
- Citations by year
- 2026: 48 · 2025: 3
- FWCI
- 48.02
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- neural information processing systems (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › AI-based Problem Solving and Planning
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Information Systems › Information Retrieval and Search Behavior
- Keywords
- agentic memory, LLM agents, knowledge networks, memory organization, dynamic indexing, evolution of memory, knowledge representation, memory management
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-09-28, under version 3 of the claim definition (prompt extract-v3).
A second model labelled each claim on 2026-09-28 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.